Llama-31-8B_task-3_120-samples_config-3

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B-Instruct on the GaetanMichelet/chat-60_ft_task-3 and the GaetanMichelet/chat-120_ft_task-3 datasets. It achieves the following results on the evaluation set:

• Loss: 0.4408

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 8
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 150

Training results

Training Loss	Epoch	Step	Validation Loss
2.3729	1.0	11	2.4972
2.6938	2.0	22	2.4571
2.6474	3.0	33	2.3881
2.2763	4.0	44	2.2642
2.0268	5.0	55	2.0694
1.7309	6.0	66	1.7871
1.4481	7.0	77	1.4330
1.0554	8.0	88	1.0675
0.8392	9.0	99	0.7563
0.4685	10.0	110	0.6437
0.3588	11.0	121	0.5851
0.6319	12.0	132	0.5407
0.4211	13.0	143	0.5248
0.495	14.0	154	0.5127
0.4232	15.0	165	0.5019
0.496	16.0	176	0.5103
0.3903	17.0	187	0.4814
0.331	18.0	198	0.4913
0.2403	19.0	209	0.4869
0.3563	20.0	220	0.4718
0.4107	21.0	231	0.4596
0.2631	22.0	242	0.4478
0.4212	23.0	253	0.4496
0.3304	24.0	264	0.4408
0.3296	25.0	275	0.4437
0.3266	26.0	286	0.4441
0.1403	27.0	297	0.4496
0.1732	28.0	308	0.4574
0.1797	29.0	319	0.4809
0.1355	30.0	330	0.4990
0.1346	31.0	341	0.5313

Framework versions

• PEFT 0.12.0
• Transformers 4.44.0
• Pytorch 2.1.2+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1